Reasoning Runtime Behavior of a Program with LLM: How Far Are We?